Correlation‐Preserving Interpolation Schemes for multi‐stripe analysis

Two alternative approaches for densifying the Multi‐Stripe Analysis (MSA) results are presented, aiming to obtain synthetic MSA data at intensity measure (IM) levels not originally captured by the analysis. Motivation comes from the computational cost of using complex structural models to assess the seismic response, often leading to running MSA at few IM levels with a limited number of ground motions. Still, a continuous representation of multiple engineering demand parameters (EDPs) given the IM level is sometimes required, for example, for rapid damage assessment when sensor information is available after a seismic event that corresponds to an “intermediate” IM value not matching the ones available from MSA. The first proposal is for a non‐parametric method that employs a correlation‐preserving record‐wise linear interpolation scheme. The second is a parametric method that employs a correlation‐preserving moment‐wise linear interpolation scheme based on a joint lognormal distribution assumption. The validity of the proposed methods is examined against actual MSA data calculated for a nine‐story steel moment resisting frame building and a reinforced/prestressed concrete bridge. Both approaches show merit, subject to sensitivity to sample size: As expected, interpolation fidelity degrades when the spacing between the IM levels increases and/or fewer ground motions are available.